Our long-term goal is to help elucidate mechanisms of temporal lobe epilepsy and develop more effective treatments for patients. During the previous funding period we discovered increased activity of dentate granule cells beginning minutes before spontaneous seizures in a rat model of temporal lobe epilepsy. We now propose to further evaluate the entorhinal-hippocampal circuit to localize the site(s) of earliest preictal activity (Aim 1), test potential mechanisms of seizure initiation (Aim 2), and determine whether unit recording can be used to predict seizures in real time (Aim 3). Aim 1 is to use tetrodes to record unit activity in dentate gyrus, CA1, CAS, subiculum, and medial entorhinal cortex (layers II, III, and V/VI) in epileptic pilocarpine- treated rats as they experience spontaneous seizures. Electrographic seizure onset will be standardized by objective methods applied to field potential recordings from dentate gyrus in all cases. Neuron firing rates will be plotted with respect to seizure onset and analyzed to determine whether and when they significantly exceed baseline levels. The region(s) displaying earliest preictal unit activity will be identified as a potential site of seizure initiation. Aim 2 is to begin testing hypotheses of seizure initiation: that in epileptic rats granule cells fail to provide sufficient excitatory synaptic drive to GABAergic interneurons (Experiment 2a) and that within the aberrant, recurrent, excitatory dentate circuit granule cells with basal dendrites are hyper- connected elements and therefore potential initiators or propagators of seizure activity (Experiment 2b). In collaboration with John Huguenard we will use laser scanning photo-uncaging of glutamate to focally activate granule cells in hippocampal slices while obtaining whole cell recordings of evoked excitatory postsynaptic currents in interneurons (Experiment 2a) and granule cells (Experiment 2b), which will be labeled with biocytin to determine whether they have basal dendrites or not. Aim 3 is to develop classifier algorithms and use on-line multi-unit recording and analysis to predict spontaneous seizures of epileptic pilocarpine-treated rats in real time. Broadband recordings of prolonged periods of interictal and preictal activity will be obtained, evaluated, and made available to other investigators. The proposed experiments will help reveal underlying causes of temporal lobe epilepsy by beginning to localize the brain region(s) displaying earliest pre-seizure activity and test potential mechanisms of seizure initiation. In addition, we will attempt for the first time to predict seizures by monitoring action potential firing rates. If successful in epileptic rats, this method may eventually lead to seizure prediction devices for patients.